Re: [OpenBIOS] r255 - in openbios-devel: arch/ppc/qemu include/openbios modules

[Blue Swirl]

On 11/24/08, [e-mail]  wrote:
> Author: stepan
>  Date: 2008-11-24 13:21:36 +0100 (Mon, 24 Nov 2008)
>  New Revision: 255
> void
>  -nvram_init( void )
>  +nvram_init( char *path )

Bad idea, this breaks Sparc32 build.

arch/sparc32/openbios.c: In function 'arch_init':
arch/sparc32/openbios.c:121: error: too few arguments to function 'nvram_init'
make[1]: *** [obj-sparc32/target/arch/sparc32/openbios.o] Error 1

r255 - in openbios-devel: arch/ppc/qemu include/openbios modules

[Laurent Vivier]

Le lundi 24 novembre 2008 à 18:59 +0200, Blue Swirl a écrit :
> On 11/24/08, [e-mail]  wrote:
> > Author: stepan
> >  Date: 2008-11-24 13:21:36 +0100 (Mon, 24 Nov 2008)
> >  New Revision: 255
> > void
> >  -nvram_init( void )
> >  +nvram_init( char *path )
>
> Bad idea, this breaks Sparc32 build.

I've looked in the device tree of sparc32 and didn't find any entry for
nvram, so is the call to nvram_init() really needed ???

Regards,
Laurent

r255 - in openbios-devel: arch/ppc/qemu include/openbios modules

[Blue Swirl]

On 11/25/08, Laurent Vivier  wrote:
> Le lundi 24 novembre 2008 à 18:59 +0200, Blue Swirl a écrit :
> > On 11/24/08, [e-mail]  wrote:
> > > Author: stepan
> > >  Date: 2008-11-24 13:21:36 +0100 (Mon, 24 Nov 2008)
> > >  New Revision: 255
> > > void
> > >  -nvram_init( void )
> > >  +nvram_init( char *path )
> >
> > Bad idea, this breaks Sparc32 build.
>
> I've looked in the device tree of sparc32 and didn't find any entry for
>  nvram, so is the call to nvram_init() really needed ???

Yes, not for registering it to the device tree, but for reading the
config from nvram. The same applies to Sparc64.

The function could be changed to check for a NULL (passed to it by
Sparc32/64) and perform different actions then, or maybe nvconf_init
should be exported and Sparc changed to use it instead of nvram_init.

r255 - in openbios-devel: arch/ppc/qemu include/openbios modules

[Laurent Vivier]

Le 25 nov. 08 à 17:40, Blue Swirl a écrit :
> On 11/25/08, Laurent Vivier  wrote:
> > Le lundi 24 novembre 2008 à 18:59 +0200, Blue Swirl a écrit :
> > > On 11/24/08, [e-mail]  wrote:
> > > > Author: stepan
> > > > Date: 2008-11-24 13:21:36 +0100 (Mon, 24 Nov 2008)
> > > > New Revision: 255
> > > > void
> > > > -nvram_init( void )
> > > > +nvram_init( char *path )
> > >
> > > Bad idea, this breaks Sparc32 build.
> >
> > I've looked in the device tree of sparc32 and didn't find any entry  
> > for
> > nvram, so is the call to nvram_init() really needed ???
>
> Yes, not for registering it to the device tree, but for reading the
> config from nvram. The same applies to Sparc64.
>
> The function could be changed to check for a NULL (passed to it by
> Sparc32/64) and perform different actions then, or maybe nvconf_init
> should be exported and Sparc changed to use it instead of nvram_init.

IMHO the best thing to fo is to export and call nvconf_init.

Laurent

----------------------- Laurent Vivier ----------------------
"The best way to predict the future is to invent it."
- Alan Kay

r255 - in openbios-devel: arch/ppc/qemu include/openbios modules

[Andreas Färber]

Laurent,

Am 24.11.2008 um 17:59 schrieb Blue Swirl:
> On 11/24/08, [e-mail]  wrote:
> > Author: stepan
> > Date: 2008-11-24 13:21:36 +0100 (Mon, 24 Nov 2008)
> > New Revision: 255
> > void
> > -nvram_init( void )
> > +nvram_init( char *path )
>
> Bad idea, this breaks Sparc32 build.

[snip]

Thanks *a lot* for reviving your old ppc patches!

They certainly weren't a bad idea in general, just them too quickly  
being applied without further review.

Andreas